[svlug] time slows down.

[Rafael Skodlar]

HW clock is only used during bootup to give the OS initial time. After
that you can still access it but that's not necessary since the OS keeps
it's own track of time. When you connect to NTP server, you update your
OS clock which is the one that matters. Once a day should be sufficient
for most users.

date shows system clock while hwclock shows hardware clock. Either tool
can be used to udjust time in it's domain.

It's a good idea to update HW clock after NTP updates the OS clock every
once in a while, more often if you reboot daily. Since the PC is up, the
HW clock get's it's power from the main power supply so the battery
doesn't matter. However, you might have some strange hardware that
somewhat depends on good clock battery rather than PSU so you better
check it out. It's also possible that the battery is hidden inside a
large chip which you need to replace alltogether.

On Tue, Oct 15, 2002 at 12:18:58PM +0200, Ira Abramov wrote:
> yesterday I noticed my server's clock was way behind, which is odd since
> I've always ran NTP on it and it was always correct. NTP indeed showed a
> big offset fronm the two ntp servers I point to. I ran ntpdate and after
> 2 hours there was again a huge drift.
> 
> this morning I stopped NTP (6 hours were missing), deleted the drift
> file, deleted /etc/adjtime (just in case), ran ntpdate, made sure I get
> the correct time (compared to my NTP-synched workstation), wrote it to
> the hwclock, and now look what happens, without NTP running:
> 
> green:/etc# date;hwclock -ru
> Tue Oct 15 11:55:36 IST 2002
> Tue Oct 15 11:55:35 2002  -1.042848 seconds
> green:/etc# date;hwclock -ru
> Tue Oct 15 11:56:18 IST 2002
> Tue Oct 15 11:56:44 2002  -2.430042 seconds
> green:/etc# date;hwclock -ru
> Tue Oct 15 11:56:23 IST 2002
> Tue Oct 15 11:56:53 2002  -3.537605 seconds
> green:/etc# date;hwclock -ru
> Tue Oct 15 11:56:32 IST 2002
> Tue Oct 15 11:57:02 2002  -0.999015 seconds
> green:/etc# date;hwclock -ru
> Tue Oct 15 11:56:37 IST 2002
> Tue Oct 15 11:57:08 2002  -0.147153 seconds
> green:/etc# date;hwclock -ru
> Tue Oct 15 11:56:43 IST 2002
> Tue Oct 15 11:57:19 2002  -0.925445 seconds
> green:/etc# date;hwclock -ru
> Tue Oct 15 11:56:48 IST 2002
> Tue Oct 15 11:57:26 2002  -0.152801 seconds
> green:/etc# date;hwclock -ru
> Tue Oct 15 11:57:06 IST 2002
> Tue Oct 15 11:57:46 2002  -0.271383 seconds
> green:/etc# date;hwclock -ru
> Tue Oct 15 12:02:55 IST 2002
> Tue Oct 15 12:05:41 2002  -0.548752 seconds
> green:/etc# date;hwclock -ru
> Tue Oct 15 12:06:12 IST 2002
> Tue Oct 15 12:10:18 2002  -0.639831 seconds
> 
> 4 minutes lost in 15 minutes.
> 
> the machine is a P2/500, not SMP, running the same 2.4.18 with gentoo
> patches I run on most of my machines, Debian Woody.
> current load average is 1.66, 1.44, 1.39 but most of the day it's pretty
> close to 0 (just a few private pages and a small mail server). it's been
> up for 43 days.
> 
> anyone got any idea what causes the kernel to lose system time like
> this? anything to try before I compile a vanilla kernel and reboot?
> 
> -- 
> Ira Abramov | Etagon | 054-731-963
> 
> _______________________________________________
> svlug mailing list
> svlug at lists.svlug.org
> http://lists.svlug.org/lists/listinfo/svlug

-- 
Rafael
The Gap Between the Rich and the Poor is Constant.